HOXA9 Acts As a Regulatory Switch in Acute Myeloid Leukaemia And
Total Page:16
File Type:pdf, Size:1020Kb
bioRxiv preprint doi: https://doi.org/10.1101/2021.03.05.434116; this version posted March 17, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. 1 2 HOXA9 acts as a regulatory switch in acute myeloid 3 leukaemia and myeloproliferative neoplasms 4 Laure Talarmain1, Matthew A. Clarke2, David Shorthouse3, 5 Jasmin Fisher2, Benjamin A Hall3* 6 1. MRC Cancer Unit, University of Cambridge, 7 Hutchison/MRC Research Centre, Box 197, Cambridge 8 Biomedical Campus, Cambridge, CB2 0XZ, United 9 Kingdom 10 2. UCL Cancer Institute, Paul O'Gorman Building, 72 11 Huntley Street, 12 London, WC1E 6BT, United Kingdom 13 3. Department of Medical Physics and Biomedical 14 Enginering, Malet Place Engineering Building, 15 University College London, Gower Street, London 16 WC1E 6BT, United Kingdom 17 *to whom correspondence should be addressed 18 [email protected] 19 The authors declare no competing financial interests. 20 1 bioRxiv preprint doi: https://doi.org/10.1101/2021.03.05.434116; this version posted March 17, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. 21 Abstract 22 Blood malignancies arise from the dysregulation of 23 haematopoiesis. The type of blood cell and the specific order 24 of oncogenic events initiating abnormal growth ultimately 25 determine the cancer subtype and subsequent clinical 26 outcome. HOXA9 plays an important role in acute myeloid 27 leukaemia (AML) prognosis by promoting blood cell expansion 28 and altering differentiation; however, the function of HOXA9 in 29 other blood malignancies is still unclear. Here, we demonstrate 30 the importance of this gene in chronic myeloproliferative 31 neoplasms (MPN) and highlight the biological switch and 32 prognosis marker properties of HOXA9 in AML and MPN. This 33 binary switch function can explain the clinical stratification of 34 these two blood disorders. First, we establish the ability of 35 HOXA9 to stratify AML patients with distinct cellular and 36 clinical outcomes. Then, through the use of a computational 37 network model of MPN, we show that the self-activation of 38 HOXA9 and its relationship to JAK2 and TET2 can explain the 39 branching progression of JAK2/TET2 mutant MPN patients 40 towards divergent clinical characteristics. Finally, we predict a 41 connection between the RUNX1 and MYB genes and a 42 suppressive role for the NOTCH pathway in MPN diseases. 2 bioRxiv preprint doi: https://doi.org/10.1101/2021.03.05.434116; this version posted March 17, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. 43 Introduction 44 Blood cancers are malignancies that can arise from any type of 45 blood cell and dramatically affect haematopoiesis. 46 Myeloproliferative Neoplasms (MPNs) are chronic diseases of 47 the myeloid lineage characterised by an excessive production 48 of fully functional terminally differentiated blood cells. These 49 have been classified into 3 types: polycythemia vera (PV), 50 essential thrombocythemia (ET), and primary myelofibrosis 51 (PMF) [1]. Despite the relatively good prognosis of these 52 diseases, MPN patients are at high risk of thrombosis and can 53 develop a blast phase MPN (MPN-BP) [2]; a subtype of the 54 blood cancer Acute Myeloid Leukemia (AML) with poor survival 55 outcomes [3]. The frequency of MPN transformation to blast 56 phase MPN is highly related to the initial MPN disease type [4, 57 5,6]. Therefore, a better understanding of the molecular events 58 driving the different subtypes of MPNs is essential to help 59 diagnose patients with higher risk of thrombosis and AML 60 progression. 61 62 AML itself is an aggressive blood and bone marrow 63 malignancy defined by the uncontrolled growth of myeloid 64 progenitor cells along with a myeloid-lineage differentiation 65 arrest [7]. As with MPN, there exist different types of AML with 66 a broad range of morphologic, cytogenic and immunologic 67 features, all associated with diverse clinical outcomes [8]. 68 Despite their similarities; prognosis, symptoms, and genetic 69 alterations differ between AML and MPN. For example, JAK2 3 bioRxiv preprint doi: https://doi.org/10.1101/2021.03.05.434116; this version posted March 17, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. 70 mutation is the main driver event of MPN diseases yet is rarely 71 found in de novo AML [9]. However, myeloid lineage 72 dysregulation by both MPN and AML, as well as the ability of 73 MPN to evolve to AML, indicate that both diseases may share 74 similar biological mechanisms. The identification of these 75 processes will help identify aberrant genes and pathways in 76 blood malignancies that could be targets for new drugs. 77 78 Better understanding of the patterns of genetic alterations in 79 cancer cells can also be used for the classification of blood 80 diseases and prediction of progression into more severe forms 81 of the disease [10]. How different combinations and orders of 82 mutations lead to different subtypes of cancer remains a major 83 open question [11, 12]. The importance of mutation order has 84 been demonstrated in MPN by Ortmann et al [13], who show 85 that two subpopulations of patients with MPN can be 86 distinguished by the order of mutation acquisition between the 87 TET2 and JAK2 genes and that these subpopulations have 88 distinct clinical characteristics. Further analyses of these 89 cohorts show that patients with JAK2 mutated before TET2 are 90 younger at presentation of the disease in clinics, are more 91 likely to present with PV, have a higher risk of thrombosis and 92 respond better to JAK2 inhibitor ruxolitinib. However, the 93 molecular interplay between both mutations within cancer cells 94 and how their order rather than their combination triggers 95 dissimilar clinical characteristics have not been investigated. 96 4 bioRxiv preprint doi: https://doi.org/10.1101/2021.03.05.434116; this version posted March 17, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. 97 Overexpression of a single homeobox gene, HOXA9, has been 98 reported as sufficient to quickly induce myeloproliferation, 99 gradually followed by AML progression after a period of time 100 [14]. Homeobox genes or HOX genes were first identified in 101 the fruit fly Drosophila melanogaster as essential regulators of 102 early embryogenesis [15] and are thought to have a critical role 103 in cancer development [16]. In the HOXA family, HOXA9 is the 104 most described gene in literature and its expression was 105 shown to be the single most highly correlating factor, out of 106 6817 genes tested, for poor prognosis in AML [17]. The 107 importance of HOXA9 in AML has been widely explored; 108 however, this has mainly focused on specific AML subtypes 109 such as MLL-rearranged leukaemia [18] and NUP98-HOXA9 110 induced leukaemia [19], while its role in other blood 111 malignancies such as MPN or other AML subtypes is poorly 112 characterised. Recently, the oncogenic property of HOXA9 has 113 been associated with its self-positive feedback loop in myeloid 114 precursor cells as a result of its ability to bind its own promoter 115 [20]. We hypothesise in this work that this specific property can 116 help stratify patients with blood cancers affecting the myeloid 117 lineage. 118 119 Using public datasets from AML patients and MPN studies, we 120 show that bimodal HOXA9 expression identifies two distinct 121 cohorts of patients/mice, reflecting the protein acting as a 122 binary switch in the cell. We show that this switch-like 123 behaviour enables a clinical stratification in these blood 124 diseases leading to separation of individuals into two clinically 5 bioRxiv preprint doi: https://doi.org/10.1101/2021.03.05.434116; this version posted March 17, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY 4.0 International license. 125 distinct populations. This leads to distinct prognoses, but also 126 allows specific disease type classification. First, HOXA9 127 bimodal expression affects the clinical features, such as age 128 and white blood cell counts, but also patient classification into 129 specific French-American-British (FAB) or molecular subtypes. 130 Next, we design a computational network model that offers a 131 mechanistic explanation of the distinct clinical features of MPN 132 progression in patients with different orders of JAK2 and TET2 133 mutations. This computational model predicts that HOXA9 is 134 directly downstream of JAK2 and TET2 and effectively stores 135 their mutational history, leading to a phenotypic switch in 136 double mutant cells dependent upon mutation order and 137 producing distinct subtypes of the disease. Finally, the network 138 model also predicts a suppressive role for the NOTCH 139 pathway in MPN and a new interaction between RUNX1 and 140 MYB.